Research with retarded and nonretarded populations suggest that toys can be an important part of group-care programs. Yet in a review of the literature. Wheman (1976) showed that there was a paucity of empirical data on effective toy utilization with retarded populations. In this study time-sampling measurements were made of the free-play behavior of 11 severely retarded females with 20 different toys and comparison items. Data analysis revealed that (a) the subjects had strong preferences among toys; (b) there was a low, statistically nonsignificant, correlation between toy preference and price; and (c) professional staff were unable to make accurate predictions of toy preference. Subjects were idle 65 percent of the time when only the 10 less popular toys were available, but only 25% of the time when only the 10 more popular toys were available. We concluded that (a) the behavior of retarded individuals is strongly influenced by the particular toys available to them, (b) empirical data are necessary to make maximally effective and economical use of entertainment materials, and (c) the current lack of such data for retarded population makes this an important area for future research.